AI Workflows for Biopharma with Alex Telford

What It Covers

Alex Telford, founder of six-month-old Convoke, explains how LLMs now make it feasible to automate biopharma commercial assessments — competitive intelligence, revenue forecasting, and indication selection — tasks previously requiring $300/hour consultants and months of manual Excel and PowerPoint work.

Key Questions Answered

• •Early Commercial Analysis: Biopharma companies routinely delay commercial assessments until Phase 2 or later, then discover they selected the wrong subpopulation, comparators, or indication — too late to course-correct. Running automated commercial analysis at the preclinical stage, when development paths still branch freely, prevents costly late-stage pivots and improves trial design decisions before resources are committed.
• •Indication Prioritization at Scale: Standard practice narrows 100 potential indications down to 5 for detailed review, leaving 95 unexamined. Automating competitive landscape synthesis, epidemiology pulls, and pricing benchmarks makes it feasible to analyze all 100 at low cost, surfacing non-obvious market gaps that a single analyst working manually would never reach within a project budget.
• •Forecasting as Process, Not Output: The precise revenue number in a forecast — whether $1B or $2B — matters less than the structured thinking the forecasting process forces. Building a forecast compels teams to define patient subtypes, identify required efficacy benchmarks, map competitive line-of-therapy dynamics, and clarify what clinical profile is actually needed to capture meaningful market share.
• •Capital-Constrained Strategy: Small biotechs operating with limited runway should not optimize for maximum NPV or largest addressable market. Binary survival dynamics mean prioritizing probability of success, creating value inflection points at discrete milestones, and monetizing those milestones to extend runway — a fundamentally different decision framework than large pharma portfolio management, where expected-value calculations across many assets apply.
• •Two-Cultures Problem in Pharma Software: The intersection of people who deeply understand pharma workflows and people who build high-quality software is extremely small. Pharma-native founders build products with poor engineering; tech-native founders target visible drug discovery problems and miss operational white space — areas like CRO invoice auditing, competitive intelligence automation, and revenue forecasting — that only become visible after years working inside the industry.