Automations
Create and manage GovTribe AI automations that can start from one or more schedules, government-record changes, or workspace events, with optional email notifications.
Click Automations, or open GovTribe AI and choose Automations.
Automations let GovTribe AI run from a saved configuration instead of waiting for you to start every prompt manually. Use automations for recurring research, saved-search monitoring, government-record changes, pipeline review, pursuit follow-up, and other repeatable AI work. You can also choose whether GovTribe emails you when an automation run finishes.
Each automation has one or more start triggers. Every trigger is an independent OR condition: a schedule, record change, or workspace event can start the same saved prompt without waiting for the other triggers to happen.
Connected AI clients can also manage automations through GovTribe MCP. Use Create_Automation to create a new automation, Search_Automations to find an existing automation, and Update_Automation to change an automation.
Requirements
Automations require available credits, a project, and a prompt. If credits are not enabled or available for your account, GovTribe may send the account owner to Credits before a new automation can be created.
| Requirement | What it means |
|---|---|
| Credits | GovTribe requires available credits to create or update automations. |
| Project | Each automation belongs to a project. Generated conversations from the automation also belong to that project. |
| Prompt | The prompt tells GovTribe AI what to do each time the automation starts. |
| Start triggers | One or more schedules or supported events control when the automation runs. |
Projects are required because automations create ongoing work. Keeping them in a project makes the generated conversations easier to find later.
Create an automation
From Automations, choose View templates, Create via chat, or Create manually from the Create via chat menu.
View templates opens a starter library for common capture, proposal, market intelligence, and pricing automations. Choosing a template starts a GovTribe AI setup conversation for that template. Create via chat opens GovTribe AI with a setup prompt that can inspect your saved searches, pipelines, pursuits, projects, recent AI work, and buyer or vendor activity before recommending high-value automations. Create manually opens the direct automation form.
If you are using GovTribe MCP, call Create_Automation with the automation name, prompt, a non-empty start_nodes list, and GovTribe AI project name. To change an existing automation, find its current version and start-node IDs with Search_Automations, then call Update_Automation with the expected version number and the fields or start-node operations you want to change.
| Setting | What it controls |
|---|---|
| Name | The automation name shown in the Automations list and search results. |
| Project | The project where the automation and generated conversations belong. |
| Prompt | The instruction GovTribe AI runs each time the automation starts. |
| Reasoning level | How much deliberate reasoning GovTribe AI uses during automation runs. Medium is the recommended default for new automations. |
| Skill | The GovTribe AI skill used for the automation, when selected. |
| Connected tools | External connectors available to the automation when connector access is enabled and the provider is connected. Use Manage connected tools to connect or review the available tools first. |
| Run when | The first schedule or event that can start the automation. Choose + Also run when… to add more independent start triggers. |
| Schedule fields | The recurrence, day, start time, minute, timezone, custom recurrence rule, and Exact or Approximate timing policy for scheduled runs. |
| Event fields | The government record, saved search, pipeline, pursuit movement, stage, or direction details for event-triggered runs. |
| Saved-search frequency | How often a saved-search trigger checks for new results: Instant, Daily, or Weekly. This is separate from the saved search's email cadence. |
| Email me | Whether GovTribe AI emails you after every run, never, or only when the completed run matches your instructions. |
The automation prompt composer is focused on the saved automation instruction. It is not the same as sending a normal conversation message.
Choose a reasoning level
Reasoning level is saved with each automation and applies whenever that automation runs, including Run now tests. Existing automations keep their saved level. New automations use Medium when you do not choose another level.
| Level | Use it when |
|---|---|
| None | The work is mechanical, such as scanning for a clearly defined condition, copying fields, or applying an explicit rule. |
| Low | The automation needs routine judgment, classification, summarization, or a straightforward fit check. |
| Medium | The automation must compare several signals, resolve ambiguity, or make a consequential recommendation. This is the recommended starting point. |
| High | The work requires difficult multi-step analysis, competing evidence, or substantial planning. Use it selectively. |
Higher reasoning can take longer and use more credits. Start with Medium for most automations. Choose None or Low for clearly mechanical or routine work, and raise the level only when the automation needs more judgment.
The automation setting is independent of your Tune Intelligence preference. Automated turns currently run on GPT-5.6 Terra, and the saved reasoning level governs the automation run and its tool work. If you open the generated conversation and send a follow-up message yourself, that ordinary turn uses your current Tune Intelligence and Response length preferences. See Manage conversations.
From GovTribe MCP, send reasoning_level as none, low, medium, or high. Create defaults to medium when the field is omitted. Update preserves the current value when the field is omitted.
Combine multiple start triggers
Configure the first Run when… row, then choose + Also run when… for each additional schedule or event. All rows run the same prompt, use the same project and skill, and follow the same completion email setting.
| Goal | Start triggers to combine |
|---|---|
| Watch one solicitation for amendments without losing a weekly review cadence | A government record changes for the opportunity OR a weekly schedule. |
| Track a likely recompete and discover a separate follow-on notice | A government record changes for the award or IDV OR a saved search that finds new related opportunities OR a schedule for broader research. |
| Watch any of five specific records | Add one A government record changes trigger for each record. |
| Review capture work after movement and on a regular cadence | Pursuit moves stages OR a daily or weekly schedule. |
The automation starts when any one configured trigger fires. It does not wait for all triggers to match, and separate triggers can create separate runs. When you choose Run now on an automation with several triggers, use Run as if to select the start condition the test should simulate.
Choose completion email settings
Automations can email you after a run finishes. Email is the built-in completion notification: it goes to the user who owns the automation and includes a link back to the generated automation conversation. A saved prompt can also explicitly send an attributed message to another owned GovTribe AI conversation, as described below; that is a conversation-tool action rather than a completion callback.
| Email me setting | What it does |
|---|---|
| On every run | Tells GovTribe AI to send an email after every automation run finishes. Use the optional text box to tell GovTribe AI what the email should always include. |
| When something specific happens | Tells GovTribe AI to send an email only when the completed run matches the condition you describe. Use the text box to explain what should trigger the email. |
| Never | Does not send an automation email. |
The text box can also include instructions about the email content, such as what to highlight, how to summarize the result, or what tone to use. For When something specific happens, the text box is required because it defines the condition GovTribe AI evaluates before emailing you.
From GovTribe MCP, use the flat on_completion_notify_user key with always, conditionally, or never. Use completion_notification_instructions for the condition and email-writing guidance. Do not send internal notification_node or notification_config keys from MCP. When using Update_Automation, omit both completion email fields to keep the current setting, include them to change it, or set on_completion_notify_user to never to turn completion emails off.
Monitor or message another conversation
Every automation run is a separate GovTribe AI conversation in the automation's project. A run does not execute inside the conversation where you created it, subscribe to another transcript, or automatically resume that conversation when it finishes.
When Full Conversation Recall is enabled, the saved prompt can read visible messages from a known owned conversation. Without recall, the run can inspect metadata and working status but cannot inspect that conversation's transcript. A reliable monitor keeps the monitored conversation ID separate from the delivery conversation ID, stores a last-observed message ID or timestamp, and sends nothing when the observed output is unchanged.
If the saved prompt explicitly sends a message to another owned conversation, GovTribe records Sent from another GovTribe AI thread on the delivered message and links back to the authorized source. Sending can start an idle destination or steer an active one. It is an explicit conversation update, not an automatic callback, join, or subscription.
Use schedules
Scheduled automations can run once at a future time, on a common recurring pattern, or on a custom recurrence.
| Schedule | Use it when |
|---|---|
| Hourly | The automation should check or summarize work throughout the day. |
| Daily | The automation should run once per day. |
| Weekdays | The automation should run on business days. |
| Weekly | The automation should run on a selected day each week. |
| Custom | You need a RFC 5545 RRULE recurrence pattern outside the standard schedule choices. |
Choose Exact when GovTribe should add no deliberate load-spreading delay. Exact does not guarantee wall-clock execution: GovTribe checks due schedules once per minute, and queue pickup, worker capacity, outages, and task duration can still delay a run.
Choose Approximate when a short forward-only load-spreading delay is acceptable. The automation may start at the selected time or shortly after. New schedules use Approximate by default, and you can change the timing policy while creating or editing an automation.
GovTribe MCP schedule triggers use the same policies through jitter_policy in trigger_config: standard is Approximate and remains the default, while none is Exact and stores the next run at the canonical occurrence without adding deliberate jitter.
Custom schedules use RRULE syntax. The selected timezone determines the local time for scheduled runs.
| RRULE | Schedule it creates |
|---|---|
RRULE:FREQ=DAILY;COUNT=1;BYHOUR=9;BYMINUTE=0;BYSECOND=0 | Runs once at the next 9:00 AM occurrence. |
RRULE:FREQ=DAILY;BYHOUR=8;BYMINUTE=0;BYSECOND=0 | Runs every day at 8:00 AM. |
RRULE:FREQ=WEEKLY;BYDAY=MO;BYHOUR=9;BYMINUTE=30;BYSECOND=0 | Runs every Monday at 9:30 AM. |
RRULE:FREQ=MONTHLY;BYDAY=MO,TU,WE,TH,FR;BYSETPOS=1;BYHOUR=9;BYMINUTE=15;BYSECOND=0 | Runs on the first weekday of each month at 9:15 AM. |
Use COUNT=1 for a one-time scheduled run. After the final occurrence is claimed, a schedule-only automation remains published but appears as Completed with No upcoming runs. If the automation also has an event trigger, it remains Active because that event can still start it; the completed schedule is omitted from the upcoming-run summary.
Changing a schedule's RRULE or timezone gives it a new starting point and rearms it. Editing the prompt, project, reasoning level, notification settings, or another unrelated field preserves a completed schedule and does not replay it. Pausing and resuming also do not replay a completed schedule.
Use event triggers
Event-triggered automations run after GovTribe records a supported government-record or workspace event.
| Trigger | What can start the automation |
|---|---|
| A government record changes | GovTribe records new activity for one selected government record. |
| Saved search gets new results | A selected saved search receives new matching results at the automation trigger's selected frequency. |
| Pipeline created or updated | A selected pipeline changes, or any accessible pipeline changes when no pipeline is selected. |
| Pursuit created or updated | A pursuit in the selected pipeline changes. |
| Pursuit moves stages | A pursuit moves between supported stages, optionally scoped by direction or stage configuration. |
GovTribe adds trigger and run context to the automation so the generated conversation can explain what started the run.
Set a saved-search trigger frequency
Every Saved search gets new results trigger requires its own Instant, Daily, or Weekly polling frequency. Instant triggers check every 15 minutes. Daily and Weekly triggers check on their corresponding cadence.
When you first select a saved search while creating an automation, GovTribe copies an existing Instant, Daily, or Weekly saved-search frequency. If no supported frequency is available, the builder starts with Daily. This is a one-time default that you can change before saving; it is not a continuing link between the two settings.
Later changes to the saved search's email frequency do not change the automation trigger, and turning saved-search email off does not disable the automation. The automation's completion email setting is also separate from both frequencies.
A new or retargeted saved-search trigger begins listening when you save it. A paused, broken, or otherwise inactive automation has no pending saved-search check. When the automation becomes active, GovTribe schedules the next check from that time using the trigger's frequency.
Resuming an automation also restarts listening from the resume time, so results from before that point are not replayed. Editing an active automation preserves saved-search progress when the target stays the same, then schedules the next check from the time you save the new version.
Monitor a government record
Choose A government record changes, select a Record type, and then search for the exact Government record to monitor. A new or retargeted trigger begins listening when you save that change. Resuming the automation restarts listening from the resume time, so earlier activity is not replayed.
You can monitor one record from any of these families with each trigger:
- Federal Contract Award
- Federal Contract IDV
- Federal Contract Opportunity
- Federal Contract Vehicle Opportunity
- Federal Grant Opportunity
- Federal Grant Award
- Federal Forecast
- State/Local Contract Opportunity
This trigger is useful for watching an opportunity for amendment-driven deadline, description, or link changes; watching an award or IDV for modifications and lifecycle changes that affect recompete timing; or watching a grant opportunity or forecast as its details develop. GovTribe can group several newly recorded changes for the selected record into one automation run.
The trigger watches only the exact selected record. It does not automatically start when GovTribe adds a separate recompete, follow-on opportunity, related award, or other linked record. For broader follow-on discovery, add a saved-search or schedule trigger to the same automation. See Monitor pursued opportunities and follow-ons for a complete monitoring pattern.
Example automation templates
Use the template library when you want GovTribe AI to help shape the prompt, start triggers, and email behavior from a common starting point. Templates are still normal automations after setup; you can tailor the saved prompt, project, trigger scopes, and email condition before GovTribe creates anything.
| Template | Recommended trigger | Email me setting | Good fit |
|---|---|---|---|
| Saved-search opportunity triage | Saved search gets new results | When something specific happens | Rank new saved-search matches against your fit profile and separate pursue, partner, follow-up, and weak matches. |
| Qualify new matches and prepare outline-ready bids | Saved search gets new results | When something specific happens | Gate new saved-search matches through bid/no-bid decisions and only outline targets that meet your threshold. |
| Executive gate-review email | Pursuit created, pursuit updated, or pursuit moves to Qualification/Review | When something specific happens | Produce an executive-ready bid/no-bid recommendation without creating proposal artifacts. |
| Recurring capture monitor and pipeline report | Weekly, daily, saved search gets new results, or pipeline updated | On every run | Send a recurring delta report for a market, saved search, pipeline, recompete lane, or buyer lane. |
| Pipeline health review | Weekly schedule or pipeline/pursuit updated | On every run | Review stale pursuits, missing values, weak stage fit, linked-record gaps, and unclear next actions. |
| Pursuit stage-change briefing | Pursuit moves stages | On every run | Brief the capture team when a pursuit enters or exits a stage and call out deadline risk or missing context. |
| Recompete and follow-on monitoring | Weekly schedule or saved search gets new results | When something specific happens | Watch for likely follow-ons, incumbent clues, and timing risks that become actionable in a capture window. |
| Specific award recompete watch | Weekly or monthly schedule | When something specific happens | Monitor one award, IDV, vehicle, contract number, or GovTribe record for follow-on activity and timing changes. |
| Buyer or agency market pulse | Weekly schedule or saved search gets new results | On every run | Track material changes in a buyer lane's buying model, set-aside posture, vehicles, value bands, and live demand. |
| Partner, incumbent, and competitor watch | Weekly schedule or saved search gets new results | When something specific happens | Identify likely bidders, incumbents, competitors, and partner candidates worth capture-team action. |
| Solicitation compliance triage | Saved search gets new results, pursuit updated, or pursuit moves to Proposal/Review | When something specific happens | Extract deadlines, instructions, evaluation factors, required content, amendments, Q&A, and compliance risks. |
| Pricing evidence pulse / pressure alert | Weekly schedule, pursuit updated, or pipeline updated | When something specific happens | Watch pricing evidence for material posture changes, strong comparables, or evidence that changes price-to-win assumptions. |
Manage automation runs
Automation list and detail views can show the automation status, trigger summary, project, next run while active, last run, run history, and current notification setting. Generated conversations can be opened from their run history when available.
Automations are edited by the person who owns them, and each saved change creates a new version. GovTribe saves your edit against the version you opened. If the automation changed in the meantime — for example, you saved an edit in another browser tab, or a connected AI client updated it for you — your edit is not applied. GovTribe warns you that the automation changed and asks you to reload it, so you can review the current setup before making the change again. GovTribe also asks you to reload and try again if another save for that automation is still finishing. An automation you have deleted can no longer be updated.
Connected AI clients follow the same rule, which is why they look an automation up before changing it. A change applies only the settings and triggers included in the request and leaves the rest of the automation as it was. See Update automation MCP tool for the exact fields.
| Action | What it does |
|---|---|
| Run now | Starts the automation manually. For multiple start triggers, choose which condition the test should simulate. |
| Pause | Stops future automatic runs and removes any pending saved-search check without deleting the automation. |
| Resume | Turns a paused automation back on and schedules saved-search checks from the resume time. |
| Retry | Runs a failed automation attempt again when retry is available. |
| Edit | Updates the automation configuration. |
| Delete | Removes the automation after confirmation. |
Automation status shows whether an automation is active, paused, broken, scheduled to run soon, complete, failed, or otherwise unable to run. A published schedule-only automation appears as Completed with No upcoming runs after all of its finite schedules finish. If an automation is missing a project or marked broken, review it before expecting configured triggers to start new runs.
Related articles
- Create a conversation: Start the project conversations that automations can work alongside.
- Choose a skill: Choose the right skill for an automation prompt.
- Create automation MCP tool: Create automations through GovTribe MCP.
- Update automation MCP tool: Change existing automations through GovTribe MCP.
- Search automations MCP tool: Find automations and recent runs before updating them.
- Automation data type: Review fields returned for automation records.
- Reasoning level attribute: Review the stable values accepted by automation APIs and MCP tools.
- Monitor pursued opportunities and follow-ons: Combine exact-record changes with saved searches and schedules for broader follow-on discovery.
- Manage connected tools: Connect official tools, control chat tool selection, and request a custom connector.
- GovCon workflows with GovTribe AI: Use detailed workflow guidance and copyable prompt examples.
- Tune GovTribe AI: Manage global Intelligence and Response length preferences, memories, conversation recall, GovTribe data access, and custom instructions.
- Saved searches: Create saved searches that can trigger automations.