Tabs and Campfire teamed up to host the Future Finance Forum at SF Tech Week 2026. Bringing together hundreds of B2B finance leaders and founders to discuss where AI is taking the future of finance.
Check out the onstage sessions below to hear how industry leaders are approaching the finance function while usage based pricing, agents, and token cost are revolutionizing the workstream.
Usage-Based Pricing: What CFOs Need to Get Right
Fireside Chat with Ali Hussain (Tabs CEO) and Jesal Bhuta (RSM principal).
Ali Hussain and RSM's Jesal Bhuta discuss why pricing is shifting from seats to usage and outcome, how rethink your finance stack to support everything from product metering and sales motion to contracts and revenue recognition, and set up systems that support today's model while you transition.
Key takeaways
- Tech spend has jumped. Tech and cloud spend used to run about 3% of human capital costs and it’s now jumped to 20% to 50%.
- Plan for 16 types of contract motions. An RSM study found co-terming, early renewals, and similar motions each hit billing, revenue recognition, and entitlements differently.
- Enterprises want annual commitments, not overage. Most buy a bundle upfront and rarely agree to automatic upgrades or overage. Vendors usually absorb overage costs until renewal.
- Start small and keep platforms flexible. Pilot first, build a phased roadmap with one joint team, and avoid point solutions that won't evolve with you.
Building the Modern Finance Stack
Controller Panel with Kat Sands (Campfire), Mark Longaroux (Assort Health), and Greg Capitolo (Attivo Partners)
The panel shares how they're using AI agents to close faster and scale without adding headcount. They walk through what they've automated, where a human still needs to review the work, the skills finance teams need now, and what auditors expect when AI touches the numbers.
Key takeaways
- Zero-day close is possible for small companies. For companies under $5 million in revenue with uncomplicated contracts, Greg's team expects a zero-day close once the right platforms and agents are in place.
- Hire later, and map the workstream first. The first finance hire used to come around 25 employees, and it's now closer to 50 or even 100 because agents do the work. To shorten new hire onboarding, map the agentic process end to end, implement systems, and close the books a few times.
- Agents draft, people review, and auditors expect the same controls. Agents can prepare work papers, draft non-recurring journal entries, and create a GAAP financial statement workbook, but controllers still review every output. Agentic errors are usually misinterpreted questions rather than hallucinations, which is why people need to stay in the loop.
Real Cost of Scaling AI: APIs, Agents, and the New Finance Stack
Keynote with Arjun Gopalratnam (Tabs CPO)
Arjun Gopalratnam, Tabs' Chief Product Officer, walks through what scaling AI really costs, from rising token bills to the organizational changes that follow. He predicts flatter teams, faster feedback loops, and paying top performers enough to keep them to offset climbing AI spend, diverging models, & the required review of AI-generated work.
Key takeaways
- Budget for 50% AI cost inflation, and match models to jobs. Token spend is climbing about 50% year over year and expects that to repeat in 2027. Model prices are also diverging, with some recent models costing 10 times as much as lower-end ones, so match each model to the job instead of defaulting to the top tier.
- Review is the bottleneck, so measure outcomes. Creating content has gotten cheap, but reviewing it hasn't, so plan for more review layers and don't assume a polished document is correct. Tracking lines of code or documents produced rewards the wrong behavior, so track the business results the work drives instead.
- Flatten the team, and pay top performers to stay. Organizations will shift from a pyramid to a column, with fewer junior roles, more senior people, and feedback happening daily instead of quarterly. Show employees what their equity is worth using real numbers, and give strong performers more scope and autonomy.
Data Accuracy in the Age of AI: Month-End Close and Audit Readiness
Panel with Brendan Doyle (Head of Product, Campfire), Arjun Gopalratnam (CPO, Tabs), and Lisa Ricardo (VP Enterprise Performance & Technology, CBIZ), moderated by Nick Jezienicki, (Partnerships, Campfire)
Brendan Doyle (Campfire), Arjun Gopalratnam (Tabs), and Lisa Ricardo (CBIZ) discuss where new risk shows up when review moves from manual to automated, and how to stay audit-ready when AI touches most of the numbers before a human does.
Key takeaways
- Separate deterministic from non-deterministic work, and control both. Use LLMs to pull information from text, like contract terms, and use deterministic software for math. When numbers go wrong, it's usually the inputs or prompt, not the math. Put change controls on prompts and agents like any other software.
- Keep a human in the loop, and check exceptions all month. Small items the AI passes as immaterial can add up to something material, so review them and use exception control logs continuously, not just at month-end. Create separate agent teams that can also check each other by using different models and flagging any disagreement.
- Match autonomy to your stage, and prove it first. Large SEC filers should watch AI closely, while cash-basis startups can let agents handle bank reconciliations now. Run an AI-built accrual process side by side with humans for about six months before setting guardrails.
- Fix the source data before automating. Use deterministic rules first, then LLM reasoning on top for cash applications.
