Every entry has a source
Memory comes only from real conversation turns. Each addition references the turn it came from, with an exact short excerpt.
Discovery call assistant
Discovery calls where the details don’t slip.
Decide what you need to learn, such as volumes, units, baselines, constraints, and decision steps. The coach tracks it from the buyer’s own words and suggests the next useful question.
Mac app · One-time purchase · No meeting bot Uses your OpenAI API key. API usage is billed separately.
Call memory, mid-callFictional example
- Volumes and units
- “About 1,200 active SKUs across two warehouses” · Buyer
- Current baseline
- “Stockouts on top sellers roughly twice a month” · Buyer
- Constraints
- “Two planners, no dedicated analyst” · Buyer
- Next steps
- Data review proposed, not yet agreed · Seller
Suggested question
Who else would weigh in before a pilot budget is approved?
The hard part of discovery is keeping track
Good discovery depends on specifics: how many, how often, compared with what, and who decides. On a live call you are listening, thinking about the next question, and trying to remember the number the buyer mentioned twenty minutes ago. Details slip, questions get asked twice, and the follow-up email is missing the figure that mattered.
AI Sales Call Coach keeps a compact memory of the call in fields you define. It also uses that memory to shape cues, so the next suggestion builds on what you already know instead of repeating discovery.
Call memory
Memory fields you define
Each memory field has a name and a plain-language instruction describing what belongs in it. Add, edit, or switch off fields for each business profile. Some starting points, all illustrative:
Technical evaluation
Environment and systems · Volumes and units · Performance baseline · Hard constraints · Open technical questions · Agreed actions
Services and consulting
Scope · Budget range · Timeline · Stakeholders · Success measures · Already answered
Founder-led SaaS
Current workaround · Problem and impact · Decision process · Concerns · Next steps
Grounded
Built from what was actually said
Speakers stay attributed
A buyer requirement is recorded as the buyer’s requirement. It isn’t treated as proof that you can meet it.
Proposals aren’t commitments
A next step you proposed stays separate from one the buyer agreed to, so your notes reflect where things actually stand.
Suggestions aren’t evidence
Cues, your feedback, and reviews are never memory sources. Only what you and the buyer said counts.
Exact numbers, recalled
When the coach recalls a constraint, it uses the actual quantities and units rather than “your limits”.
No repeated questions
The coach checks whether a question was already answered and avoids reopening settled topics.
After the call
Everything you learned, saved on your Mac
Each session keeps the speaker-labeled transcript, the cues you saw, and a copy of the profile you used. Generate an editable review that picks out the most useful observation from the call and suggests better wording where it would help. Add your own notes. Export the session as JSON if you want to move details into your CRM or notes, or delete it when you no longer need it.
Want to practice a technical discovery first? Run a typed AI sales roleplay. The AI buyer has a backstory you uncover by asking good questions.
FAQ
Discovery call questions
Does it integrate with my CRM?
No. Export a session as JSON to move details into your CRM or notes.
Can I use different memory fields for different kinds of calls?
Yes. Memory fields belong to a profile, so create one profile per type of call.
Does memory keep growing as the call gets longer?
No. The coach works from the recent conversation plus compact memory, so it stays focused on long calls. The full transcript is still saved.
Apple silicon · macOS 14.2+ · Requires an OpenAI API key; API usage is billed separately by OpenAI.