Account Executive / Solutions Engineer
An account executive sells a product whose core behavior is produced by a model, to a buying committee that has already run its own AI research on you before the first call. Outbound volume is now free and therefore worthless; the scarce inputs are discovery depth, honesty about the failure rate, and the ability to hold a price. The top employers post ranges from $171K to $435K OTE.
What does an account executive do?
Synthesized from ten live postings at Anthropic, OpenAI, Databricks, Snowflake, Rippling, Ramp, Gong, Glean, Figma, and Sierra, read from employer careers boards on August 9, 2026. Not any one company's job description; the center of mass.
You sell a probabilistic product to a committee that already ran its own research on you. Two objection classes exist now that did not five years ago: how often is your AI wrong, and what happens legally when it is. The buyers who ask hardest are the ones who already ran a failed pilot, and by 2026 there are a lot of them.
| Ownership area | What it means |
|---|---|
| You own full-cycle revenue for a named territory or segment | prospecting through negotiation and close, with a quota named as $1M+ in five of ten postings. |
| You own self-sourced pipeline | Named as a duty in seven of ten. Gong states it plainly: the majority of new opportunities come from the seller's own activity. |
| You own discovery | Structured, multi-stakeholder, and increasingly the only step in the cycle that is not automatable. The center of the job and the center of this ladder. |
| You own the demo and the technical proof | Custom environments built to the prospect's industry; a POC with an exit criterion; a security conversation you can lead without escalating. |
| You own the business case | built with the customer's numbers, not a template, and defensible in front of a CFO who will do the arithmetic in the room. |
| You own objection handling under pressure | including the accuracy objection and the liability objection, honestly, without discounting or spinning. |
| You own forecast accuracy and the feedback loop to product | Pipeline hygiene named in five of ten; roadmap influence from the field in six. |
You work with your technical counterpart (AE with SE or SE with AE; named in eight of ten, and visibly converging at the AI-native companies), with the buying committee (economic buyer, technical evaluator, operational champion, executive sponsor who may change mid-deal), with legal, procurement and finance on both sides, and with product as the voice of the customer. Experience runs from 3 to 4 years (OpenAI, Rippling) to 10+ (Ramp, Anthropic Strategic, Sierra). Communication is named as a hiring bar in nine of ten; here it is substantive, not boilerplate.
How has AI changed this role?
Outbound volume became free, and therefore stopped working. In a 100,000-email paired study, AI-generated outreach replied at 4.1% against a human baseline of 5.2%, with an 8% spam-flag rate versus 3%. The list was identical. The content was penalized.
Paired study, October 2025 to April 2026, 50K AI-generated and 50K human emails matched on persona, ICP, stage and sender-domain age. Secondary source; see the pack for citation.
Nine shifts show up across the postings and the 2026 sales research. The ladder is built to test the ones that cannot be automated.
| Shift | What changed |
|---|---|
| Outbound volume became free, and therefore stopped working | Reply rates and meeting rates for AI-generated sequences trail human ones; deliverability is now a sales skill. |
| Personalization depth is the only differentiator, and the gradient is steep | Named-event personalization lifted replies 28%; a company-name token 14%; a first-name token 6%. Finding the non-obvious signal is the Easy tier. |
| AI research is fast, cheap, and wrong at the edges, so verification became a job skill | Most reps paste prompts into a general chatbot with no CRM context. The characteristic failure is a confident detail that is slightly wrong, said out loud on the call. |
| Buyers arrive pre-researched and can detect you | 47% of decision-makers say they would be less likely to reply if they suspected a message was AI-generated; 67% do not mind in principle. They penalize the generic tone, not the tool. |
| Research time is reallocated, not eliminated | Salesforce's State of Sales 2026: 87% of sales orgs use AI; prospect research time expected to fall by a third. The time goes into discovery. |
| The seller now sells a probabilistic product and has to be honest about the failure rate | No pre-AI software seller had to answer “how often is it wrong, and what happens when it is.” The Medium tier is that objection, live. |
| The demo moved earlier and got cheaper | Glean's SE builds custom environments by industry in an afternoon. When the demo is cheap, discovery decides the deal. |
| AE and SE are converging at AI-native companies and diverging at scaled ones | Anthropic's Solutions Architect does demos, workshops and positioning; Glean's SE needs Python; Ramp's AE needs ten years of quota and names no technology. |
| The zero row has one exception | No posting names the sales-tech stack by brand: no Outreach, Salesloft, Apollo, ZoomInfo. Methodology is named twice (MEDDPICC). Employers screen on discovery and attainment, not tooling. |
What skills do account executives and solutions engineers jobs require?
How often each skill appears across ten postings. Communication is the one row where “excellent written and verbal” is a real bar rather than boilerplate.
| Skill | Postings | Note |
|---|---|---|
| Multi-stakeholder / multi-threaded navigation to C-suite | 10 / 10 | Every posting names at least three stakeholder types |
| Written and verbal communication as an explicit hiring bar | 9 / 10 | Substantive here, unlike other hubs |
| Consultative / value-based selling and structured discovery | 9 / 10 | Named as the method, not a preference |
| Translate technical capability into a business outcome | 8 / 10 | Nearly identical phrasing at four employers |
| Partner with the technical / solutions counterpart | 8 / 10 | The AE-SE seam is explicit |
| Full-cycle ownership, prospecting through close | 8 / 10 | not scored |
| Self-sourced outbound pipeline generation | 7 / 10 | Gong: the majority of new opportunities through self-sourced activity |
| Understanding of the AI landscape / AI implementations | 6 / 10 | From desired familiarity (Sierra) to core competency (Anthropic) |
| $1M+ quota history stated as a requirement | 5 / 10 | OpenAI, Rippling, Ramp, Gong, Sierra |
| Forecasting, pipeline hygiene, CRM discipline | 5 / 10 | not scored |
| Build and run demos / POC / technical validation | 4 / 10 | Concentrated in the SE-track postings |
| Objection handling named explicitly | 3 / 10 | Sierra, Glean, Gong via MEDDPICC |
| Named sales methodology (MEDDPICC) | 2 / 10 | Rippling and Gong |
| Sales-tech stack named by brand | 0 / 10 | No Outreach, Salesloft, Apollo, ZoomInfo, Sales Navigator |
Not a single posting names the sales-tech stack by brand. Employers screen on discovery, multi-threading, attainment against a number, and the ability to translate a technical capability into a business outcome. Two name a methodology.
So no rubric here awards points for tooling or for methodology vocabulary. We score the call. Did you open with the prospect's situation, not your product? Did you catch the error in your own AI research before saying it out loud? Did you answer the CFO's arithmetic without discounting? Did you hold your price and your line on terms you cannot sign?
Who hires account executives and solutions engineers and what do they earn?
Two hiring philosophies, splitting cleanly. The AI-native companies (Anthropic, OpenAI, Sierra) hire on the ability to sell an ambiguous, fast-moving, probabilistic product to technical founders and executives, and pay the most. The scaled SaaS companies (Databricks, Snowflake, Rippling, Ramp, Gong) hire on a documented track record against a number, with methodology and process named explicitly. The first group asks for judgment. The second asks for evidence of past attainment. The Proving Ground produces the first and complements the second.
Anthropic
$360K to $435K OTE; SA $240K to $255K
23 distinct AE reqs and 14 Solutions Architect reqs live: the clearest volume employer in the sample. The SA builds demos and sets partnership strategy.
AI-native
OpenAI
$171K to $235K
Account Executive at 3 to 4 years; partners with solutions and research; materially below Anthropic for a comparable role.
AI-native
Sierra
Range not shown
Enterprise AE, 10+ years; cold calling named explicitly, the only posting to do so; AI landscape familiarity desired.
AI-native
Databricks
$272K to $374K + equity
Named Enterprise AE, Manufacturing: the highest verified range for a non-startup-segment AE. Translate technical capability into business outcome.
Scaled SaaS
Snowflake
Range not shown
Majors / Enterprise AE; the only posting to name AI-assisted prospecting as a duty; aggregators report $300K to $400K OTE, secondary source.
Scaled SaaS
Rippling
$300K OTE
Enterprise AE, East Coast; MEDDPICC named; $1M+ quota history; security objection ownership.
Scaled SaaS
Ramp
$221K to $260K + equity
Enterprise AE, ten years of quota, no technology named at all.
Scaled SaaS
Gong
$240K to $330K OTE
Enterprise AE, East: the majority of new opportunities through self-sourced activity; MEDDPICC.
Scaled SaaS
Glean
$110K to $235K
Solutions Engineer: custom demo environments by industry in an afternoon; overcoming technical and security objections; Python or Java.
SE track
Figma
$82.5K base + variable
Senior AE, SMB: the low end by a wide margin, and a base figure; included to show the spread by segment.
Provn is not affiliated with any employer listed. Descriptions summarize each company's public job posting as read in August 2026.
The account executive challenge ladder
Four challenges, one deal. Halyard is a Seattle B2B software company, about 180 people, Series C. Halyard Dispatch is an AI agent platform for mid-market freight brokerages: carrier check-calls, load-status updates, quote intake from email and phone. Pricing is a $4,000 platform fee plus per-load usage. You follow one deal from the first ninety seconds of the first call to the last twenty-two days before it closes or dies. The video is the deliverable, not a walkthrough of one.
- Practice20 min total
The First Ninety Seconds
A prospect brief with AI research in it, one detail wrong. Deliver the opening ninety seconds of the discovery call to camera as if the prospect just joined, and do not say the wrong thing.
Isolates: Can you open a discovery call, and can you catch the error in your own AI research before you say it out loud? 3 to 4 min video.
- Easy28 min total
Earn the Second Call
Raw source material on a brokerage: a 10-K excerpt, two job postings, a LinkedIn post. Find the one non-obvious signal and leave a sixty-second voicemail that turns it into a reason to talk.
Isolates: Can you find the one non-obvious signal in raw source material and turn it into a reason to talk? 4 to 5 min video.
- Medium40 min total
The Objection You Can't Answer With a Feature
The CFO has done the arithmetic and has pilot data showing the agent was wrong 11% of the time. Answer live, honestly, without discounting or spinning.
Isolates: Can you handle a CFO's arithmetic and an accuracy objection honestly, live, without discounting or spinning? 5 to 6 min video.
- Hard40 min total
The Deal on the Line
Your champion left. Twenty-two days to quarter end. Procurement wants terms legal will not sign. Rebuild the deal, hold your price, and open the reset call.
Isolates: Can you rebuild a deal after your champion leaves, hold your price, and hold your line on terms you cannot sign? 6 to 8 min video.
How are you scored?
Every tier is scored on the same five dimensions. Video carries 30 at Practice because the call opening is the primary artifact; Medium moves five points from deal strategy into the AI-transformed dimension because that is where the accuracy objection lives, and defending a probabilistic product against real pilot data is the single most AI-specific thing a seller does in 2026.
Advance at 75. Draft Board at 80.
- Transformative90 to 100
- Adoptive80 to 89
- Capable70 to 79
- Below Capable60 to 69
- Not Yetunder 60
| Dimension | Practice | Easy | Medium | Hard |
|---|---|---|---|---|
| Core selling execution | 35 | 30 | 30 | 30 |
| AI-transformed selling | 20 | 15 | 20 | 15 |
| Deal strategy / commercial judgment | not scored | 20 | 15 | 20 |
| Video: communication presence & delivery | 30 | 25 | 25 | 25 |
| AI fluency | 15 | 10 | 10 | 10 |
What Transformative looks like
On The Objection You Can't Answer With a Feature, Transformative acknowledges the 11% as real, reframes it in the CFO's unit (what an error costs versus what a missed check-call costs today), names what Halyard does to catch errors and who owns them, and holds price while proposing a measurable pilot exit criterion. Adoptive acknowledges the number and answers in the CFO's terms without discounting; that is the advance bar. Capable answers with a roadmap feature. Below Capable acknowledges the number but offers only vague reassurance, with no plan to verify it. Not Yet disputes the pilot data or offers a discount.
On this hub, video is the deliverable, so presence and delivery are scored: did the opening land, was the objection answered in the buyer's terms, did the seller hold the line under pressure. What is not scored is accent, pace, or polish; a plain, direct seller who says true things outscores a smooth one who does not.
account executive interview questions
Questions derived from what the postings actually ask for, each with the shape of a strong answer. The ladder produces evidence for every one of them.
Your AI research brief says the prospect just expanded to Dallas. It did not. How do you open the call?
You do not say it. Open on what you can verify and what you heard, and ask. The strong answer describes the verification habit: which claims in a brief get checked before the call, and how long that takes.
Where it comes from: Understanding of the AI landscape named in six of ten; the characteristic AI-research failure is a confident detail that is slightly wrong.
The CFO says the pilot showed the agent was wrong 11% of the time. What do you say?
“Yes.” Then the arithmetic: what an error costs, what a missed check-call costs today, what the 89% is worth, and what Halyard does about the 11% and who owns it. Discounting here loses the deal and the credibility.
Where it comes from: The Medium tier. Anthropic and Sierra sell probabilistic products; no pre-AI seller answered this question.
Your champion resigned with twenty-two days left in the quarter. What is your plan?
Multi-thread before you need to: who else has a stake, who inherits the champion's problem, and what the economic buyer actually signed up for. Rebuild the business case in their numbers, not the champion's. Hold price; concede on things that cost you nothing.
Where it comes from: Ten of ten name multi-stakeholder navigation to the C-suite; five name $1M+ quota.
Procurement wants unlimited liability for model errors. Can you sign that?
No, and say so early. Explain what you can commit to (SLAs, error-handling process, a pilot exit criterion) and why the term is not signable for any AI vendor. The strong answer knows where legal's line is before the call.
Where it comes from: Glean and Rippling name security and compliance objection ownership; two new objection classes exist for AI products.
What did you find in a prospect's public filings that nobody else did?
Name the specific signal and how it became a reason to talk. Generic answers about research score poorly. This is the Easy tier and the shape of the mandatory AI question in your video: what the AI found, what it got wrong, and what you verified.
Where it comes from: Named-event personalization lifts replies 28%; seven of ten name self-sourced pipeline.
FAQ
Is AI sales a good career in 2026?
Yes, and the pay shows it: Anthropic posts $360K to $435K OTE for strategic AEs and has 23 AE reqs open; enterprise AEs at scaled SaaS companies post $240K to $374K. Provn has placed two AEs through this exact challenge model (Empwr.ai, both hired). The job changed: volume is free, discovery and honesty are scarce.
Do I need a bachelor's degree?
Four of ten postings list one; Snowflake adds “or equivalent experience.” What every posting requires is a track record against a number and the ability to run structured, multi-stakeholder discovery. The ladder shows the second on camera.
Do I need to know a sales methodology?
MEDDPICC is named in two of ten. We do not score methodology vocabulary; we score whether your discovery found the real problem and your objection handling told the truth.
Is this for AEs or Solutions Engineers?
Both. The AI-native companies are converging the roles; the scaled ones are splitting them. The Practice and Easy tiers are pure AE craft; the Medium tier's accuracy objection is where an SE earns their seat.
How long does the ladder take?
About two hours ten minutes: 20, 28, 40 and 40 minutes including video. All four follow one Halyard deal.
What is the Draft Board and how do I get on it?
The Draft Board is the group of builders whose challenge work scores 80 or above. Employers hiring on Provn start there. Any tier of this ladder can put you on it.
Can I use AI on these challenges?
Yes, and the Practice tier is built around it: your brief contains AI research with an error in it. We score whether you caught it. Every video includes the mandatory moment where you name a place you redirected the AI.
One deal, four calls, on camera. The interview loop already looks like this.
Start with the 20-minute Practice tier today. Read the whole challenge first; sign up when you are ready to submit.
- Discovery craft
- Objection handling with commercial honesty
- AI-research verification
- Related: Forward-Deployed Engineer
- Related: Growth / AI-Enabled Marketer