Getting Started with Agents

Last updated: April 9, 2026

What are Agents?

Agents is one of the newer features within Glyphic and it is evolving the most quickly — changes are being made week to week. Agents let you ask questions and run analyses across multiple accounts, deals, or conversations at once, going far beyond what Ask Glyphic can do at a single call or deal level.

Agents vs. Ask Glyphic

Ask Glyphic

Agents

Best at the call, company, or deal level

Expansive — spans multiple accounts, deals, and CRM data

One-off, in-context queries

One-off prompts or scheduled recurring runs

Manual, on-demand

Can be triggered by deal stage changes or run on a schedule

No CRM context beyond linked records

Access to CRM fields, stages, and deal values — availability varies by scope (deal vs. company level)

How to Run a One-Off Agent

  1. Navigate to the Agents page in Glyphic.

  2. Enter your prompt in natural language. Examples:

    • "Across my open deals, what are the biggest risks?"

    • "Which of my customers are showing signs of churn risk? Identify critical events and suggest mitigation steps."

    • "In Josh's deals, has a champion been identified in every deal?"

  3. Glyphic will ask a series of clarifying questions to refine the search — answer as many or as few as you like.

  4. The output appears on a subsequent screen. You can edit it before sharing or saving.

Web Search Toggle

When you enter a prompt you will see a Web Search toggle. Turn this on when you want external data — for example, competitor analysis after a discovery call to understand similar companies, their pain points, or what other tooling they use.

CRM Context

Agents can also reach into your CRM via your tracked fields. Any field you have mapped in Glyphic from Salesforce or HubSpot becomes additional context the agent can use when crafting its analysis.

Recurring Agents

Instead of running a prompt manually each time, you can set up a recurring agent in two ways:

  • Deal-stage triggered — fires automatically when a deal moves to a new stage. For example, when a deal reaches Closed Won, automatically generate a handoff note for the CS team.

  • Calendar scheduled — runs daily, weekly, or monthly. At the end of every week, for instance, get a summary of all conversations across your pipeline delivered straight to your inbox or a Slack channel.

Pre-Built Templates

Scroll down on the Agents page to find the template library. Templates are pre-written prompts you can launch with one click. Categories include:

  • CRO — pipeline health and forecast summaries

  • Account Executives — Next Call Prep, Objection Predictor, Daily Call Prep

If you build a prompt you use regularly, reach out to your Customer Success Manager and we can add it to the platform catalog for your whole team.

What Data can an agent see?

We're alway continuously updating our agent capabilities, below is a list of what data our agent can currently access, and data we're looking to add in the future.

✅ - The Agent can do this

🚧 - Not Quite Yet - We're working on it!

❌ - Not supported

Data/Context

Call Level

Deal Level

Company Level

Organization Level

Call transcript

✅ Full transcript

✅ All transcripts across the deal

✅ All transcripts across the company

✅ All transcripts (across entire pipeline)

Call metadata (date, duration, participants)

✅

✅

✅

✅

Call summary

✅

✅

✅

✅

Call tags (e.g. Discovery, Demo, Closing)

✅

✅

✅

✅

Custom Insights (extracted per call)

✅

✅

✅

✅

Qualification annotations (MEDDPICC, PACT, "3 Whys")

✅

✅

✅

✅

Competitor mentions

✅

✅

✅

✅

Sentiment / engagement signals

✅

✅

✅

✅

CRM fields — Company (e.g. Hiring Status, Languages, Phone Call Tools, Qualification Framework)

❌

🚧

✅

✅

CRM fields — Deal (e.g. Deal Stage, Forecast Category, Source, Net Pipeline Impact, CRM System used)

❌

✅

🚧

✅

CRM fields — Contact (e.g. Job Title)

❌

✅

✅

✅

Deal stage, amount, close date

❌

✅

🚧

✅

Product insights (Deal Blockers, Feature Requests, Promised Features)

✅

✅

✅

✅

Conversation / relationship history

❌

✅ Full deal history

✅ Full company history

✅

Product Description (static AI context — always injected)

✅

✅

✅

✅

Organizational Memory (playbooks, stage definitions, style guides)

✅

✅

✅

✅

Custom Insight definitions (field descriptions & scoring rubrics)

✅

✅

✅

✅

Web search / external data

❌

❌

❌

✅ Toggle available on one-off agent runs

Cross-account / cross-deal analysis

❌

❌

❌

✅

Scheduled or triggered automation

❌

❌

❌

✅

While Organizational Memory is technically present in every agent's system prompt, longer or more detailed memory content may not be weighted heavily by the agent unless the prompt explicitly references it (e.g. "using our stage exit criteria defined in memory...".)

Example Use Cases

Sales

  • Pipeline risk review — "Across all my open deals, what are the biggest risks?" Great for weekly pipeline calls.

  • Champion identification — "In every open deal, has a champion been identified?" Managers can run this to surface at-risk opportunities before they stall.

  • Competitive intelligence — "Across my accounts this quarter, which competitors were mentioned most?" Use web search for enriched external context.

  • CS handoff — Triggered automatically when a deal closes to generate a structured account summary for the success team.

  • Weekly digest — Scheduled every Friday to summarise all conversations and flag next steps across the whole team.

Customer Success

  • Churn risk analysis — "Which of my accounts are showing signs of negative sentiment or reduced engagement? Rank by revenue." Schedule weekly to stay ahead of at-risk customers.

  • Renewal preparation — "For all accounts coming up for renewal in the next 90 days, summarise the relationship status, key concerns, and recommended talking points."

  • Upsell identification — "Which customers have mentioned positive sentiment and expressed interest in additional features? Identify upsell or cross-sell potential."

  • Portfolio overview — "Give me an overview of all my accounts — which are thriving, which are stable, and which need attention?"

  • Feature request tracking — "What product features or improvements have my customers requested most often across all calls this quarter?"

Tips

  • Start with a broad prompt and let the clarifying questions narrow it down — you do not need to be precise upfront.

  • For management use, scoping to a specific rep ("in Josh's deals...") or segment ("customers with revenue over $X...") gives more actionable results.

  • If you land on an agent prompt that works well for your team, share it with your CSM so it can be added to the shared template library.

  • Recurring agents are most powerful when the output goes somewhere actionable — a Slack channel or an email digest.