The future of workflow orchestration: AI agents and enterprise SaaS

Chris Pitchford is the founder of Brev, an AI-powered business performance platform that deploys agents to continuously monitor and track a company's OKRs and business goals - pulling live data from the tools teams already use (Salesforce, HubSpot, Jira, Snowflake, Slack) and surfacing risk, bottlenecks, and progress without requiring manual updates or status meetings. Before Brev, Chris spent three years scaling an OKR software company from zero to 1,200 customers - including Slack, Dropbox, Coinbase, Plaid, and Discord - before it was acquired by Microsoft in late 2021. He spent a year inside Microsoft learning what enterprise sales really looks like at scale, then left when ChatGPT launched to build the same product he'd always wanted: goal-tracking plus agents, not just dashboards.

Brev launched on Product Hunt this month and hit number one. The product is SOC 2 Type 2 compliant - unusually battle-hardened for a launch-stage product - because Chris went through full InfoSec and legal review with every design partner before writing a line of product. The team is sub-10 people, primarily engineering, and is backed by angels drawn from the customer base itself: CFOs, COOs, chiefs of staff, and VPs of Business Operations who live inside the problem every day.

What Brev Does and Who It's For

The target user is a chief of staff or VP of Business Operations at a company of 200 or more employees. This person's job is to connect strategy to execution - to make sure the plan from the C-suite actually gets done at the team level. In practice, they spend most of their time in spreadsheets and nudging people in Slack to update their goals. Brev automates that entire lifecycle.

Agents connect to each of the company's goals and spend their time - continuously - figuring out how to move that goal forward. They pull data from the relevant systems of record, flag when something is off track, and surface the context behind a red or yellow status (not just that something is behind, but why and what the blocker is). Execution team members don't have to log into a new tool - Brev reaches them in Slack, collects quick status inputs, and synthesizes them up for leadership. The chief of staff gets a real-time picture of the whole organization without the spreadsheet layer in between.

Revenue Generation Is an Organizational Problem, Not a Sales Problem

One of Chris's sharpest frameworks is the reframe of revenue as a cross-functional accountability problem. If a company's goal is to double ARR from $25M to $50M, hitting that number requires tracking churn (in Gainsight), expansions (in Salesforce), net new revenue (in HubSpot), demand and top of funnel (also HubSpot), and product feature commitments that are blocking customer expansions (in Jira or Linear). Each of those lives in a different system. No single tool answers the question: are we on track to double ARR this year?

Engineering has a direct influence on whether the expansion targets are met. Product has a direct influence on churn. The CRO can't hit the number alone - everyone has to be rowing in the same direction with visibility into the same goals. Brev's job is to synthesize those fragmented systems into a single view of whether the company is on track, and to make it effortless for execution teams to contribute context without treating status reporting as a second job.

The Coming Problem: Fragmented Agent Silos

Chris has already started thinking past the fragmented data silo problem. As AI agents take on more execution work - an AI SDR from 11x, a marketing agent from Jasper, autonomous agents across every department - each of those agents will own data and activity the same way HubSpot or Salesforce does today. If 50% of a marketing team's execution is handled by agents in three to five years, someone needs to align those agents to the same business goals as the humans working alongside them.

That orchestration layer is not going to come from IT. The CRO will still buy the AI SDR. The marketing team will still buy their own agent tools. But when the question is “are we on track to hit our goal?” the answer will have to pull from both human work and agent work simultaneously. Chris sees this as the next version of the problem Brev is already solving - and the reason the orchestration layer needs to live above all the individual tools, not inside any of them.

Design Partners Done Right: How to Validate Without Happy Ears

Chris went cold to hundreds of his ideal customer profiles before writing the first line of product. He was not just learning what features they wanted - he was filtering for the handful of people who would become real design partners. When someone qualified, he invited them into an exclusive program: six months of free access in exchange for deep integration access (Jira, Snowflake, Salesforce) and real involvement in product development. Crucially, every design partner signed terms and went through InfoSec and legal review. These were legitimate partnerships, not favors.

The discipline paid off at launch - reviewers noted the craft and thoughtfulness in the product, which was a direct result of customers building it alongside the team. For the discovery calls themselves, Chris recommends the Mom Test framework: ask questions that cannot be answered with polite agreement, and get to willingness-to-pay as early as possible. His specific technique: ask what would feel cheap, then ask what would feel prohibitively expensive. The number you want is just below the upper bound - anchored toward what the customer actually values, not the lowest price they'd accept.

Moat in the Age of Vibe Coding

Chris's answer to the vibe coding disruption question is unusually specific: the model is not the moat. Any competitor can rent the same foundation model tomorrow. What they cannot replicate is the corpus of goal and execution data Brev accumulates inside each customer's organization - a system of record no public model ever sees - and the embedded workflows that make Brev part of the muscle memory of every team without requiring a login.

Because Brev reaches users in Slack and in their existing flow of work rather than requiring them to visit a new product, the network of daily interactions compounds into training data and workflow integration that gets harder to displace over time. The outcome-based pricing model Chris originally wanted - only charge when companies hit their goals - remains an aspiration, but what he found in practice is that enterprise procurement teams need predictability. Meeting customers where they are meant a flat seat-based fee with a hybrid token model for heavy AI feature usage.

Building on Quicksand: The Early-Stage AI Founder Problem

Chris describes the first six months of building Brev as walking on quicksand. The underlying technology was moving fast enough that anything built for today felt like it might be obsolete by the time it shipped. His response was to orient the product roadmap six to twelve months out - not building for the current state of AI, but for what the infrastructure would be able to support by the time the product was in customers' hands. That shift in mental model turned the anxiety from paralyzing to useful.

On AI reliability in enterprise contexts: 99.9% accuracy is 100% wrong when the output is a sales forecast going out to 200 reps and a room full of executives. The leniency for errors in enterprise AI is near zero - and Chris is direct that this is a harder problem than most AI builders acknowledge. Getting directions to Yosemite wrong is forgettable. Getting the weekly ARR number wrong is career-defining. This is one of the core technical challenges Brev continues to work on.

Tools & Resources Mentioned

• Brev - AI agents for OKR and goal tracking, enterprise-grade; getbrev.ai. Sign up on the waitlist.
• Granola - AI note taker Chris uses daily; connects directly to OS/microphone without a Zoom bot, delivers cleaner results than Fathom or Read AI.
• Cursor - Vibe coding tool Chris uses for rapid feature prototyping.
• Vercel - Mentioned as a starting point for vibe coding deployments.
• The Mom Test - Framework for non-leading customer discovery interviews; recommended for any founder validating a product idea.
• First Round Capital - Recommended resource for design partner and early customer discovery frameworks.
• 11x / Jasper Agents - Examples of department-level AI agent tools that will create the next wave of fragmented agent silos.
• Vapi / Nik - Chris recommends the AI for Founders episode with Nik from Vapi on voice as the future communication layer.
• Salesforce, HubSpot, Gainsight, Jira, Linear, Snowflake - Systems of record Brev integrates with to synthesize cross-functional goal data.

Frameworks

FAQ