The earliest Google Books match for "I tend to agree" as a stand-alone sentence is from U.S. Congress, Joint Economic Committee, Subcommittee on Agricultural Policy, Policy for Commercial Agriculture—Its Relation to Economic Growth and Stability (December 16–20, 1957) [combined snippets]:
Senator SPARKMAN. I ask this question right now, if those who oppose the multiple parity system will say that we ought to retain a support program of some kind.
Mr. HOOD. I think our position is this. With all the surpluses we have on hand and the expanding agricultural plant which was discussed yesterday, we certainly cannot start out right now and take the props out from under the entire system. When we talk about adjustments we have to think of ways of adjusting from where we are to where we want to go.
I believe while we are making this adjustment that we would favor a chance for farmers to vote on whether they wanted a high level support with fairly limited acreage allotments or more acres and a much lower level of support. We favor a continuation of supports at levels which permit market prices to operate above supports most of the time.
As long as we have surpluses and a large plant we do not have all of the choices we might have if we were starting from scratch.
Senator SPARKMAN. What would you say, Mr. McMillen?
Mr. MCMILLEN. I tend to agree. As long as we have the current situation without an adequate solution, we should continue some degree of protection for the prices that farmers receive.
But I cannot convi[n]ce myself that we ought to maintain that protection at public expense at levels which tend to perpetuate the problems with which we are dealing. It is a little like trying to keep farmers in the buggy whip business—in a world which no longer has an adequate demand for what we are producing in those fields.
In context, Mr. McMillen is saying "I tend to agree with the response that Mr. Hood has just made." This is a perfectly coherent way to speak, and I doubt that any proficient English speaker/listener would have trouble making sense of the meaning.
Another instance occurs in Institute of Electrical and Electronic Engineers, "Improving the Strategy of Oral Presentations" (1967) [combined snippets]:
The second one is the bidder briefings, which tend to be typically a rehash of the exhibit in RFQ and suffer some of the same malady that I just ascribed to the technical planning briefings. Those who have had the prior communication with the offices and consequently are in a realistic position to compete for the job knew what was in mind before the RFQ came out and consequently don't need the briefing. The rest are wasting their time.
Comment: I tend to agree. I attend bidders' briefings periodically. The main purpose of them is to answer questions. Invariably few questions are asked and it is obvious why. Nobody wants to tip his hand. If he has any serious question, he's going to come around that corridor in XYZ building and collar the man who might be able to enlighten him; or use some other means to get his information. If it is serious, he is not going to bring it up in an open bidders' briefing with all of his competitors present.
Here again, the stand-alone "I tend to agree" means "I tend to agree with the opinion or argument of the person I just quoted or paraphrased.
A potentially more enlightening instance of phrase occurs in Lawrence Cranberg, "Ethics of Scientists," in Bulletin of the Atomic Scientists (September 1968):
Sample questions and responses [in a 1967 American Association for the Advancement of Science survey] follow:
"It has been said that science is entirely instrumental and therefore not concerned with the choice of social goals (which is an ethical or political problem). According to this view, the functions of the scientists are to seek effective and feasible means of reaching goals (chosen by authorities or democratically) and simply to point out the costs and consequences of alternative course of action."
6.8 [percent] I agree strongly. 18.5 [percent] I tend to agree. 28.6 [percent] I tend to disagree. 32.5 [percent] I disagree strongly. 2.5 [percent] I am not concerned with this question.
Gordon Allport, The Person in Psychology: Selected Essays (1968) [combined snippets] cites a survey-style assessment test that uses "I tend to agree" as one optional answer on multiple questions:
Six items were taken from Gilbert and Levinson's (1956) Custodial Mental Illness Ideology Scale (CMI). Example: "We should be sympathetic with mental patients, but we cannot expect to understand their odd behavior. a) I definitely disagree. I tend to disagree. c) I tend to agree. d) I definitely agree.
If nothing else, use of "I tend to agree" as a standard option on some surveys and survey-style tests during the 1950s and 1960s may have encouraged its use in everyday settings to express an inclination to support or approve of some proposition, without asserting wholehearted support for it—just as, for example, "none of the above" has become a kind of catch phrase outside the survey milieu. Even in survey settings, though, the clear sense of the wording "I tend to agree" is "I tend to agree with the opinion or viewpoint just expressed."