Presented by hub spots
The annual conference of Hub Spot for Marketing and Sales Professionals, the boundary took place in San Francisco this year, which includes three days of insights and events in marketing, sales, CX and AI innovation. It was a mixture of new, such as creators corners and tech stack showcase stage, and familiar, such as hub spot headquarters, spotlight product demo stage, hub spot academy labs for hand -on product learning, and participants’ favorite, brain history.
Hub Spot co -founder and CTO, Dharmesh Shah, on Wednesday, digs the opening hub spotlight, AI’s nuts, what really changes, because the Generative AI is more than just excellent independence, and it is good not to go anywhere soon.
"Is this a efficiency? Or is it an existential risk? I asked AI this question. He thought deeply about it for a while and brought the answer. Yes"
"Not everyone is essentially sold on the qualities of Ai. This is the most fruitful, and also the most controversial, technology of our time," Shah said. "Is this a efficiency? Or is it an existential risk? I asked AI this question. He thought deeply about it for a while and brought the answer. Yes"
How do you compete with AI?
6k an informal vote of respondents asked the question, "How do you compete with AI? Shah said that one -third of the respondents read, how would I compete against it, but he added, which prepares it as a zero game that is not useful to anyone.
"We should think about this as a positive amount of support," He explained. "The purpose is not to fight the machine. The goal is to build with the machine."
This is fragile because when AI’s capacity is growing on a efficiency, but AI’s learning curved letters are more linear, and thus is seen by embracing the technology that has learned everything from Shakespeare’s work all over, Redded threads are discussing that there is a hot dog.
"At the end of all this, this machine is capable of predicting such accuracy, which is a really good next word, which is not different from thinking. That’s why it feels so magical," Shah said. "It’s automatic with a PhD in everything. It is like having a thousand PhDs in your pocket. Suddenly you can write poetry. You can write prose. You can write a computer program. It is like nine in the matrix, except that ‘I know King Fu.’ ‘I know everything.’ You have the knowledge of the world in your pocket.
There are some defects. LLM is limited to the training figures given to them. Sometimes they deceive. They are frozen over time, and the state – but they learn. User conversations go into their long -term memories, including chats, resources they are asked to communicate, such as PDF or images, and tools such as Internet, database, third -party system and more.
How better use AI in our daily life
"My Pro Tip: Every time you sit on a computer to do something, try with AI first. See if it can help, don’t eliminate it."
So if AI is the actual deal, and the change is important, how should you use AI?
"My Pro Tip: Every time you sit on a computer to do something, try with AI first. See if it can help, don’t eliminate it," Shah said. "You will be surprised. When it doesn’t work, don’t think, oh, okay, it doesn’t work. Think, it doesn’t work yet. If this is an important case for you to use, leave yourself a calendar reminder to try again in three months. Remember, Ai is on a sufficientness. From now on, there is an opportunity for three months or six months, the same thing that no longer works will work."
The quality of the results you get is based on the quality of the model you used, the quality of your writing indicators, and the quality of the context that you put in the context window, otherwise known as quick engineering.
Model quality. Cambrian has exploded in the world of large language models, but Shah has recommended any of the three advanced frontier models: Openi’s GPT5, Anthropic to Claude, or Google Gemini. He added, but don’t eliminate it. Choose your choice or your company is using.
The quality of the gesture. Shah said that humans use less than 10 % of AI’s capacity. About 9595 % of the time, consumers repeat one of the handful of gestures that have worked in the majority of their use. Instead, he suggests that 60 % of the time, you use working indicators, prepare for 30 % of time repetition, or you are using now and you can see that you can improve them for better results, spend 10 % time on experience, or use AI for things you have never tried to work before. Metroompatting can be useful here – actually say to improve AI gestures
Beyond quick engineering
The context engineering is just what looks like – adding the right context to an application so that AI offers better results. Improving the quality of the context includes things like adding customized instructions, including every major LLM. It is primarily telling AI from what point of view he is working – personality, how it should be treated, how to answer, and so on. Once you provide customized instructions, it will use them for every answer, and your results will improve.
The second improvement is a recent change – MCP, or model context, a relatively new way of supplying tools to LLM, LLM. Any AI application that supports MCP can immediately and directly connect with thousands of applications that also support MCP, such as HubSpot, to bring this context directly to LLM.
To take action with AI agents
"Last year, in these bound 24, I predicted that this year would be the year of AI agents. I was wrong," Shah said. "This is not the year of AI agents. This is the decade of AI agents. We are just starting on this massive change in AI."
Last year Shah had launched Agent.AA place to discover, use and make your AI agents. Now more than 2 million people are using the platform. About 26,000 users have already created agents on the agent, including Shah, who took all the information from his keynote note and put it in the AI ​​agent, anyone can chat with him, U. e.
Team Strategy: Trage, Experience, Automatic and Measurement.
Next steps
How do you get the excitement of your new AI and bring it to your team? Shah offered a game plan, team strategy: tragedy, experience, automatic and measurement.
"The purpose behind this strategy is to go to AI under the leadership of individual bravery, where you have a very expensive, super -creative person, but translate these individual heroics into team habits," He explained. "Take these things and apply the team to go through this process again and again. We are big believers that tomorrow’s teams will be hybrid. It’s time to start. Even in a small way, start building your hybrid team."
But the most important thing is to remember that as is as smart like AI, man wins on EQ, or emotional quote. He added that marrying life experience with AI toll will improve your life and improve your experiences.
"But my challenge for you is that, don’t stop there," He said. "The future is not related to artificial intelligence. This is yours, with increased intelligence. AI is not here to take our place. This is meant to change the parts of our work that do not bring us happiness. We can focus on the remarkable wea to take over. The better AI gets, the more it lets us become human."
Sponsored articles are the content developed by a company that is either paying the post or having a business relationship with the venture bat, and is always clearly marked. For more information, contact Sales@Ventorbat.com.