Skip to content

AI Tab

Your all-in-one chat workspace: 30+ models, side-by-side answers, and deep integrations with your tools and data.

2-Minute Tour

Watch a quick overview of the AI tab—model picker, tools, and tips to get great results.

Quick start

  1. Choose a model from the selector (speed, creativity, or deep reasoning).
  2. Open Tools and pick what the model may use (Search, Files, Email, Drive, Evernote, SMS, MCP tools, etc.).
  3. Write a clear prompt (task → context → constraints → output format).

AI Page

Tip

For fast summaries, pick a sprinter. For sourced analysis and planning, pick a researcher. For brainstorming or naming, pick a creator.

Model selection

Fresh Focus AI supports over thirty models across top providers. The selector highlights each model’s strengths so you can match the engine to the job.

AI Models

  • Researchers (reasoning-heavy): best for sourced briefs, trade-offs, stepwise plans.
  • Sprinters (fast): best for summaries, checklists, short transforms.
  • Creators (open-ended/creative): best for campaigns, outlines, names, and concepting.

Note

You can switch models mid-conversation. The new model receives the same thread context so you can keep going without restating anything.

Tools (turn on what the AI can use)

Before you run a prompt, open Tools and enable what the model may call:

AI Tools

  • Search — multi-engine web lookups with de-duplication.
  • Files — read your uploaded files and write Markdown/HTML artifacts back.
  • Email / SMS — deliver full reports or short summaries to your inbox/phone.
  • Google Drive / Evernote / GitHub — publish reports where you work.
  • MCP tools — third-party capabilities (e.g., Google Calendar & Sheets via Zapier).

You can select tools in bulk or individually. The active set applies to the current prompt/thread.

Tip

Be intentional. If you only need analysis of a single file, enable Files and turn off external connectors to keep the run focused and fast.

Prompting patterns (copy & adapt)

Use this structure: Task + Context + Constraints + Output format (+ Sources).

Fast tech brief (speed) 

Summarize today’s biggest AI headline in ≤60 words and explain why it matters for product strategy.
Return 2 bullets: “What happened”, “Why it matters”. Include one link.

Deeper analysis (reasoning) 

Evaluate 3 competing approaches to retrieval for a 1M-token context application.
Compare cost, latency, and quality. Include assumptions and 3 citations.
Format as a table + key takeaways.

Use my file

Read `Quarterly_Report_2025Q2.html`. Extract revenue, gross margin, and top 3 risks.
Return a 120-word brief and a CSV table I can paste into Sheets.

Use integrations (MCP + Drive + Email)

Create a 1-page “Weekly Review” from this thread. Save HTML to Google Drive,
email me the full brief, and (if connected) add a Friday 4:00–4:30 PM Google Calendar event
titled “Weekly Review” with the Drive link.

Continue, clear, and resume

  • Continue: Ask follow-ups—your preferences (e.g., “≤60 words + business impact”) carry forward.
  • Clear: Start a fresh thread when you change topics.
  • Resume: Open the History tab to reopen any thread, compare past answers, or pick up work exactly where you left off.

See History for details.

Export & share

From the AI tab you can tell the AI to export the current result to:

  • Files — Markdown/HTML for instant preview and archiving.
  • Email — a formatted brief to your inbox.
  • Google Drive — HTML with links preserved.
  • Evernote — ENML in your chosen notebook.
  • SMS — a concise, 2-line summary to your phone.

For setup, visit Integrations. To automate, see Tasks.

Tips for higher-quality answers

  • Be specific about audience, length, structure, and must-include sections.
  • Show an example if you have one. Ask “match this style.”
  • Request sources & assumptions for research and analysis.
  • Constrain the format (table/checklist/email draft/HTML/Markdown).
  • Name your files and ask the AI to save exports with YYYY-MM-DD in the filename.

Troubleshooting

The model ignored a tool I enabled.
- Re-run with a line like: “Use the Search tool to find sources before answering.”
- Ensure only the needed tools are enabled to reduce ambiguity.

Answers feel generic.
- Add constraints and ask for assumptions, sources, or a compare & contrast table.
- Try a researcher model for depth.

It’s slow.
- Switch to a sprinter and trim the scope (≤150 words, 3 bullets, 1 link).
- Disable extra tools you don’t need.

I want to take this further.
- Save the thread, open Insights for proactive ideas based on your History, or convert the flow into a scheduled workflow in Tasks.

Privacy & safety

  • Your conversations and tool calls are saved to History under your account.
  • Exports (Email/Drive/Evernote/SMS) occur only when you authorize those integrations.
  • Health, legal, or financial content is for informational purposes only—verify with professionals.
  • History — reopen threads, compare models, and resume.
  • Insights — automatic ideas from your saved work.
  • Files — upload and preview artifacts; reference them in prompts.
  • Integrations — connect Email, Drive, Evernote, SMS, GitHub, MCP.
  • Tasks — automate research and reports on a schedule.