Dec 7, 2023 · 27m · no-priors
No Priors Ep. 43 | With Clara Shih, CEO of Salesforce AI
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of No Priors, Salesforce AI CEO Clara Shih discusses enterprise AI adoption, Salesforce's open model architecture, data readiness strategies, and the fundamental shift toward agentic, stochastic software.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 27.6% of the talking time here. How this is scored →
The hosts as informed peer 5.0 Guest teaching 3.7 Guest disagreement 0.2 The hosts pushing back 0.2
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 12:22 Nuancing the enterprise AI adoption sequence
In a very cordial interview, Clara offers her mildest pushback by gently qualifying Elad's 3-step adoption thesis, pointing out that companies need not clean all enterprise data before seeing production value in contact centers.
Hardest push from the hosts ▶ 8:32 Pressing on production adoption versus enterprise hypeElad challenges the general industry narrative by asking whether enterprise AI implementations are actually functioning at scale in production or remain speculative pilots.
Biggest teaching moment ▶ 19:15 Taxonomy of enterprise unstructured dataClara educates the hosts on enterprise data realities by breaking down how unstructured documents like PRDs require fundamentally different ingestion pipelines than noisy transcript logs.
The host holds their own ▶ 23:47 Sarah Guo's breakdown of AI unit economics and declining compute costsSarah demonstrates deep market expertise by articulating how declining token inference costs monotonically improve unit economics against high initial product value.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Clara Shih's Journey to Salesforce AI Leadership | 3 | 2 | 0 | 0 | Elad opens with a supportive career recap, asking Clara about her trajectory from Service Cloud to leading Salesforce AI. Clara provides an agreeable, narrative overview of her work with early NLP models at Hearsay and prototypes with early enterprise clients like Gucci. | |
| Salesforce's Open Model Strategy and Ecosystem Integration | 5 | 4 | 0 | 0 | Elad demonstrates solid domain knowledge by asking about internal proprietary models versus third-party LLMs and the maturity of agentic platforms. Clara delivers an extensive, informative overview of Salesforce's open architecture, Copilot Studio, and integration layers. | |
| Assessing Real Enterprise Adoption and Data Integration | 6 | 4 | 1 | 1 | Elad probes on whether enterprise AI adoption is genuine production deployment or mostly experimental pilot projects. When Elad proposes a linear three-step enterprise adoption framework, Clara mildly nuances it by demonstrating how CRM customer service teams can bypass full enterprise data overhauls with targeted RAG. | |
| Cross-Team AI Alignment and the Evolution of Software Engineering | 5 | 3 | 0 | 0 | Sarah inquires about cross-functional AI alignment across product management and engineering. Clara details the transition from hard-coded UI development to dynamic agentic generation, illustrating customer ROI through the Gucci case study. | |
| Managing Unstructured Enterprise Data and Data Privacy | 5 | 5 | 0 | 0 | Sarah raises the critical challenge of handling sensitive enterprise data and model boundaries. Clara educates the hosts on the practical taxonomy of unstructured data, contrasting high-signal PRDs with noisy conversational transcripts that require preprocessing before ingestion. | |
| The Paradigm Shift: Deterministic Versus Stochastic Enterprise Software | 6 | 4 | 0 | 0 | Elad and Sarah explore pricing models, the shift from deterministic to stochastic systems, and unit economics. Sarah displays strong technical and venture acumen when analyzing COGS amortization and value capture, while Clara complements the discussion with film production benchmarks. |
Statements from this episode (13)
Shih: Salesforce Built Early Service GPT Prototypes with Gucci
“We were working with a couple customers, including Gucci to develop very early prototypes of what now has become service GPT.”
Shih: Salesforce AI integrates internal models alongside OpenAI, Anthropic, Cohere, and Google
“So it's really a combination of using Whether it's Cogen from our research team, which is the, which powers Apex Cogen GPT that we have in, in our developer GPT, where you also fine tuning versions of that for domain specific models in customer service and for…”
Shih: Service reply recommendations are Salesforce's most popular AI feature
“So a great example of this, the most popular one is service reply recommendations for contact center agents.”
Shih: Most Enterprise Customers Are Still Experimenting with Generative AI
“Most customers are in the middle, right? They're still experimenting. They're realizing how important it is to get their data ducks in a row, and they're starting to do things like connect their Salesforce data cloud with their various data lakes.”
Shih: Data Cloud is Salesforce's fastest-growing organically developed product
“We're seeing our data cloud grow as the fastest organically developed product in Salesforce's history, and a lot of that is driven by this need for data to power AI, whether it's for training and fine tuning or for reg.”
Shih: AI agents will dramatically transform software development and UI design
“You know, as you're alluding to a lot, agents maturing and being able to do more, I think it's really going to dramatically transform the, how we approach software development, right? A lot of what was explicitly hard coded as different branching and execution…”
Shih reveals Generative Canvas, a dynamic UI prototype for Einstein Copilot
“We have this pretty awesome prototype. It's called generative canvas where, you know, as you're conversing with the Einstein co-pilot, it's kind of just popping up different components that you would need from within what you're doing.”
Shih: Generative AI reduced average customer service handle time at Gucci
“What we've seen is that the average handle time on support issues has gone down, and then, but instead of hanging up, the service advisor is able to have a deeper conversation”
Shih: Conversational unstructured data requires intensive pre-processing before AI retrieval
“There's unstructured data like a PRD or A service knowledge article where it's been written specifically with the intention of communicating a certain set of things, and you can probably assume everything in that unstructured document is important. Conversely,…”
Shih: Salesforce is building features to extract structured data from transcripts
“Within that unstructured data, of course, there's data that should remain unstructured and can be vectorized and embeddings, but there's some data, there's actually a lot of structured data in there sometimes, right? In a phone transcript that a retailer might…”
Shih: Software roles will shift to defining goals while AI executes
“There's a lot of other ones where the job of the software engineer and product manager and designer is going to shift from prescribing the how to Prescribing or describing the why and the what and the goal, and we leave it to the AI to figure out stochasticall…”