24 October 2024

Speaking to an Ai Avatar Ready to Take Your Sales Job ft 1mind CEO

2Frameworks

11Insights

Frameworks in this episode

Strategy6 steps

AI-Led Growth

Use persistent AI capacity to turn linear go-to-market growth exponential

Read →

Strategy5 steps

Dual-Audience AI Experience Design

Win adoption by creating value for buyers first and sellers second

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 2

Myth Buster12:00

Some Buyers May Prefer AI to Human Salespeople

The episode challenges the assumption that every buyer wants a human interaction. Developers and engineers may prefer immediate technical answers, while other buyers may value avoiding scheduling, sales personalities, and the social effort of managing a relationship.

AI should be offered as an option rather than imposed
Technical buyers may actively avoid salespeople
Always-on answers can outweigh an avatar's unfamiliarity
Adoption will vary between early adopters and laggards

“our goal is not to make this that you Force buyers to use this it's an option”
Amanda Kao · 12:00

“developers Engineers they don't want to talk salesperson”
Amanda Kao · 14:30

#buyer behavior#developers#sales#ai adoption

Myth Buster21:30

Top Salespeople Rarely Reuse Discovery as Well as AI Can

Amanda argues that even strong representatives struggle to retain and reuse details uncovered earlier in the same conversation. An AI can recall a prospect's markets and pain points several turns later, then connect them directly to relevant features without exceeding a fixed mental capacity.

Sales methodologies depend on reusing discovery information
Humans often apply insights only in follow-up emails
AI can recall details from earlier turns
Capacity limits worsen as representatives learn more products

“five turns later she's going to use that information to sell to you”
Amanda Kao · 22:30

“my reps are like dump trucks”
Amanda Kao · 23:00

#sales discovery#memory#sales methodology#ai

Hot Take· 1

Hot Take10:00

Buying and Selling Happen in the Conversation

Amanda argues that conversational AI can become the central interface for revenue work rather than another peripheral workflow tool. Once the AI participates in the live conversation, CRM updates, records, and battle cards become secondary outputs generated around that interaction.

The sales conversation is the core commercial event
Workflow automation should follow from the conversation
AI can use account and opportunity context in real time
Conversation memory can span sales and customer success

“our thesis is buying and selling happens in the conversation not around the conversation”
Amanda Kao · 11:30

#sales conversations#crm#revenue operations#automation

Explainer· 2

Explainer06:00

Why Photorealistic AI Avatars Still Respond Slowly

Amanda explains that photorealistic rendering requires substantially more cloud compute than the simpler avatar. OneMind sees latency as its leading technical challenge but expects optimization to make the delay unnoticeable.

Photorealistic generation uses significant H100 compute
The simpler avatar responds faster
Latency is the company's top engineering priority
Amanda expects near-human timing within months

“the photo real is basically spinning up in h100 in the cloud”
Amanda Kao · 06:00

“it is the number one thing we're building towards”
Amanda Kao · 06:30

#latency#avatars#h100#infrastructure

Explainer18:30

A Sales AI Needs a Goal More Than a Perfect Content Library

Amanda disputes the belief that companies need exhaustive documentation before deploying a sales AI. The system can start from limited high-quality material or ingest thousands of documents, but its critical instruction is a prioritized end goal rather than a deterministic sequence.

A small content base can still produce a capable AI
Large document collections do not create human-style capacity problems
Companies should prioritize authoritative knowledge
Goal-directed reasoning replaces rigid if-then scripting

“one might think that you need all the information in the world”
Amanda Kao · 18:30

“she's not set with a deterministic flow she's set with an end goal”
Amanda Kao · 19:30

#ai training#knowledge base#goal setting#sales enablement

Story· 2

Story01:00

Mindy Handles a Live Sales Call and Summons a Human Rep

The hosts test OneMind's avatar through a role-play involving discovery, a pricing request, and technical questions. Mindy presents slides, routes the buyer to an available account executive, and enters an active mode to answer questions during the call.

The avatar conducts conversational discovery
It can present contextually relevant slides
Pricing questions can trigger a human handoff
The AI can support a representative during a live call

“we can join video calls give presentations demo software and communicate across any channel”
Mindy · 02:00

“I choose slides based on the flow of our conversation and the topics we're discussing”
Mindy · 05:30

#ai avatars#sales demo#handoff#conversational ai

Story07:00

Why Amanda Kao Left the Bench to Build OneMind

After leaving 6sense and taking time off to have children, Amanda did not expect to return immediately as a founder. An AI lab approached her with technology seeking a vertical market, and she chose B2B go-to-market because it matched her expertise, network, and ambition to build a category rather than a point solution.

OneMind emerged from AI Foundation technology developed over five years
Amanda chose B2B sales, marketing, and customer success as the vertical
She rejected narrow efficiency tools
She wanted a category-scale opportunity

“I think this opportunity found me”
Amanda Kao · 08:00

“I didn't want to do anything small I wanted to play big”
Amanda Kao · 09:30

#founder story#onemind#b2b#category creation

Tool· 1

Tool25:30

Let AI Select the Sales Method for Each Moment

OneMind can be trained on methodologies such as Challenger, MEDDIC, and MEDDPICC, including a customer's proprietary approach. Amanda prefers allowing the AI to select the tactic suited to the current situation rather than constraining it to the single methodology a human team can realistically master.

The AI can learn multiple established methodologies
Enterprise customers can supply proprietary methods
Different moments call for different sales tactics
Human capacity is the reason teams often standardize on one method

“you want to apply the best meth best methodology at the right place in the right time”
Amanda Kao · 26:00

“humans have capacity limitations they can't learn 10 methods but I can”
Amanda Kao · 26:00

#challenger#meddic#sales methodology#adaptive ai

Takeaway· 3

Takeaway16:00

Buyers Tell AI What They Hide from Salespeople

Amanda says people become unusually candid when speaking with a nonjudgmental AI because they do not worry about hurting its feelings. A clone she used during fundraising elicited blunt investor reactions that made genuine interest easier to distinguish from polite deflection.

Buyers often soften rejection when speaking to people
AI reduces the social cost of saying no
Conversation transcripts can reveal true intent
Investor candor helped prioritize follow-up

“people get super transparent with the AI”
Amanda Kao · 16:00

“they just let their guard down they say exactly what they're thinking”
Amanda Kao · 16:30

#buyer honesty#fundraising#sales psychology#conversation data

Takeaway27:00

AI Could Give Every Sales Rep a Dedicated Engineer

The most immediate enterprise opportunity may be specialist coverage rather than replacing core account executives. AI can answer technical questions for junior representatives, BDRs, and commercial teams before expanding into strategic calls as a one-to-one sales-engineering resource.

Specialists are too expensive to attend every call
Junior sellers need immediate technical support
AI can improve the current AE-to-engineer coverage ratio
Humans can continue owning strategic relationships

“what if it's one to one what if the AI could be on every single call to just provide that support”
Amanda Kao · 28:30

#sales engineering#bdr#enterprise sales#specialists

Takeaway33:00

Build AI Use Cases for Today and Six Months Ahead

The hosts recommend splitting AI work between capabilities that are useful now and experiments designed for where the technology will be in six to nine months. Early experiments build infrastructure and operational understanding before improving models make the use case obvious to competitors.

Maintain practical and forward-looking AI use cases
Forecast near-term capability improvements
Experiment before quality reaches maturity
Build supporting infrastructure during the learning period

“you should have some AI use cases you're building for today but some AI use cases you're experimenting for where you believe the technology will…”
Kieran Flanagan · 33:30

“you actually really have to start to integrate this stuff into your go to market pretty rapidly”
Kieran Flanagan · 34:00

#experimentation#first mover#ai strategy#voice ai